Take it from someone with experience studying statistics and how to use them - surveys and polls as they are passed in the political sense are not representative of the general public at all. Here are things to consider:
1. A skilled statistician can make survey/poll results say anything they want and it will fool the general public 9/10 times. It may not stand up to actual academic scrutiny once people start delving into sampling data, response demographics, validity, and that sort of mess, but your average MSM consumer is just looking for confirmation bias anyway, they're not going to dig for that kind of stuff. To give an example, say I wanted to do a poll to show support for stricter gun control laws (obviously I would never do this because I am not a moron) in order to push legislation or campaign as a tough-on-guns politician etc... Obviously if I focus my sampling efforts on surveying people in an urban area with a history of liberal leaning policies I am much more likely to get the desired result than if I focus on asking that question in rural Appalachia. Sure, sample size might be something that is typically left in there or even prominently displayed to show "look how valid this is because of how many people I surveyed" but how many political surveys/polls have you seen where they actually reviewed, in detail, how many responses came from urban/rural/suburban/high-SES/low-SES/etc people, and mentioned this on the image or abstract they post with the poll? When this data is made available it's not generally included with the basic breakdown that is presented in the media - most people are not interested enough to follow the breadcrumbs and dig into this data.
2. This is more anecdotal than the last point, perhaps, but it is my experience and observation that certain people are just more likely to respond to polls in the first place. Sure there are outliers in the political context, but think about the type of person who is willing to stop what they're doing and take 5-10 minutes out of their day to respond to some stranger calling/random piece of mail or email/person on the corner. Think really hard and be honest about what that type of person probably represents. There may be a few karen types or boomers who are politically motivated enough to respond to strangers and answer polls, but in truth most people only respond to polls without monetary compensation if it's about something they care enough about in the first place. College students, wealthy urbanites, and activists of pre-existing ideological persuasion - these are the people who will reliably stop and waste their time for free answering questions to a stranger.
3. With this logic you also have to consider the type of polling done, where it's done once again, and ease of access. Your average boomer or geriatric isn't going to be online enough except for maybe browsing Facebook to come into contact with most polls or check their email regularly enough to respond, and most rural dwelling people don't go out and about enough to catch them with in-person street corner polling, so you're largely limited to mailed surveys (which most people wont fill out without financial incentive), phone calls (which most people won't answer since it's an unknown number), or door to door surveys (Which don't offer good sampling sizes and suffer limitations from the restricted scope of the surveying). So right there you're discounting huge amounts of people with the most common polling/survey methods, simply because they're not likely to even come into contact with the survey in the first plac.e